Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types
In vegetated areas, soil pH impacts plant growth, soil properties, and spectral characteristics. Remote sensing enables soil pH mapping by delivering detailed surface data, and while high-resolution satellite images show great potential in complex terrains, research in this area is still limited. Th...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-01-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/17/2/189 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832587578833895424 |
|---|---|
| author | Ziyu Wang Wei Wu Hongbin Liu |
| author_facet | Ziyu Wang Wei Wu Hongbin Liu |
| author_sort | Ziyu Wang |
| collection | DOAJ |
| description | In vegetated areas, soil pH impacts plant growth, soil properties, and spectral characteristics. Remote sensing enables soil pH mapping by delivering detailed surface data, and while high-resolution satellite images show great potential in complex terrains, research in this area is still limited. This study evaluated PlanetScope (high-resolution) and Sentinel-2 (medium-resolution) images in estimating soil pH across diverse land use types in southwestern China’s hilly areas. It examined how spectral variables from four seasonal images affect prediction accuracy. We integrated topographic and spectral variables at seven spatial resolutions (3 m, 10 m, 20 m, 30 m, 40 m, 50 m, and 60 m), using extreme gradient boosting (XGboost) for orchards, dry land, and paddy fields. We found that the models developed with PlanetScope images tended to achieve better prediction accuracy compared to those utilizing Sentinel-2 images. For each satellite, single-temporal images showed greater predictive power under each land use type. In particular, the spring spectral data showed desirable predictive performance for the orchards and the paddy fields, while the autumn spectral data contributed more effectively to the models for the dry land. Specifically, PlanetScope provided the best prediction accuracy for soil pH at 3 m resolution (orchard: <i>R</i><sup>2</sup> = 0.72, <i>MAE</i> = 0.24, <i>RMSE</i> = 0.30, <i>RPD</i> = 1.91; dry land: <i>R</i><sup>2</sup> = 0.77, <i>MAE</i> = 0.37, <i>RMSE</i> = 0.40, <i>RPD</i> = 2.09; paddy field: <i>R</i><sup>2</sup> = 0.66, <i>MAE</i> = 0.35, <i>RMSE</i> = 0.41, <i>RPD</i> = 1.71), while Sentinel-2 performed better at 10 m resolution (orchard: <i>R</i><sup>2</sup> = 0.67, <i>MAE</i> = 0.29, <i>RMSE</i> = 0.33, <i>RPD</i> = 1.75; dry land: <i>R</i><sup>2</sup> = 0.70, <i>MAE</i> = 0.39, <i>RMSE</i> = 0.47, <i>RPD</i> = 1.83; paddy field: <i>R</i><sup>2</sup> = 0.64, <i>MAE</i> = 0.34, <i>RMSE</i> = 0.42, <i>RPD</i> = 1.66). Our findings demonstrate that sensor selection, land use, temporal phases, and modeling resolution significantly impact outputs. High-resolution PlanetScope images prove effective for predicting soil pH in complex terrains. |
| format | Article |
| id | doaj-art-7d100e64250b4407a7a485b092fb2513 |
| institution | Kabale University |
| issn | 2072-4292 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj-art-7d100e64250b4407a7a485b092fb25132025-01-24T13:47:40ZengMDPI AGRemote Sensing2072-42922025-01-0117218910.3390/rs17020189Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use TypesZiyu Wang0Wei Wu1Hongbin Liu2College of Resources and Environment, Southwest University, Beibei District, Chongqing 400716, ChinaCollege of Computer and Information Science, Southwest University, Beibei District, Chongqing 400716, ChinaCollege of Resources and Environment, Southwest University, Beibei District, Chongqing 400716, ChinaIn vegetated areas, soil pH impacts plant growth, soil properties, and spectral characteristics. Remote sensing enables soil pH mapping by delivering detailed surface data, and while high-resolution satellite images show great potential in complex terrains, research in this area is still limited. This study evaluated PlanetScope (high-resolution) and Sentinel-2 (medium-resolution) images in estimating soil pH across diverse land use types in southwestern China’s hilly areas. It examined how spectral variables from four seasonal images affect prediction accuracy. We integrated topographic and spectral variables at seven spatial resolutions (3 m, 10 m, 20 m, 30 m, 40 m, 50 m, and 60 m), using extreme gradient boosting (XGboost) for orchards, dry land, and paddy fields. We found that the models developed with PlanetScope images tended to achieve better prediction accuracy compared to those utilizing Sentinel-2 images. For each satellite, single-temporal images showed greater predictive power under each land use type. In particular, the spring spectral data showed desirable predictive performance for the orchards and the paddy fields, while the autumn spectral data contributed more effectively to the models for the dry land. Specifically, PlanetScope provided the best prediction accuracy for soil pH at 3 m resolution (orchard: <i>R</i><sup>2</sup> = 0.72, <i>MAE</i> = 0.24, <i>RMSE</i> = 0.30, <i>RPD</i> = 1.91; dry land: <i>R</i><sup>2</sup> = 0.77, <i>MAE</i> = 0.37, <i>RMSE</i> = 0.40, <i>RPD</i> = 2.09; paddy field: <i>R</i><sup>2</sup> = 0.66, <i>MAE</i> = 0.35, <i>RMSE</i> = 0.41, <i>RPD</i> = 1.71), while Sentinel-2 performed better at 10 m resolution (orchard: <i>R</i><sup>2</sup> = 0.67, <i>MAE</i> = 0.29, <i>RMSE</i> = 0.33, <i>RPD</i> = 1.75; dry land: <i>R</i><sup>2</sup> = 0.70, <i>MAE</i> = 0.39, <i>RMSE</i> = 0.47, <i>RPD</i> = 1.83; paddy field: <i>R</i><sup>2</sup> = 0.64, <i>MAE</i> = 0.34, <i>RMSE</i> = 0.42, <i>RPD</i> = 1.66). Our findings demonstrate that sensor selection, land use, temporal phases, and modeling resolution significantly impact outputs. High-resolution PlanetScope images prove effective for predicting soil pH in complex terrains.https://www.mdpi.com/2072-4292/17/2/189hilly topographyextreme gradient boostingdigital soil mappingfeature importancesoil property prediction |
| spellingShingle | Ziyu Wang Wei Wu Hongbin Liu Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types Remote Sensing hilly topography extreme gradient boosting digital soil mapping feature importance soil property prediction |
| title | Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types |
| title_full | Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types |
| title_fullStr | Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types |
| title_full_unstemmed | Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types |
| title_short | Comparing Soil pH Mapping from Multi-Temporal PlanetScope and Sentinel-2 Data Across Land Use Types |
| title_sort | comparing soil ph mapping from multi temporal planetscope and sentinel 2 data across land use types |
| topic | hilly topography extreme gradient boosting digital soil mapping feature importance soil property prediction |
| url | https://www.mdpi.com/2072-4292/17/2/189 |
| work_keys_str_mv | AT ziyuwang comparingsoilphmappingfrommultitemporalplanetscopeandsentinel2dataacrosslandusetypes AT weiwu comparingsoilphmappingfrommultitemporalplanetscopeandsentinel2dataacrosslandusetypes AT hongbinliu comparingsoilphmappingfrommultitemporalplanetscopeandsentinel2dataacrosslandusetypes |